The present invention relates to producing micro-libraries composed of micro-colonies, arrayed on a surface, which can comprise from one to several thousand or more cells. The present invention also relates to methods and means for replicating a micro-library via formation of a replicate, "daughter cell," micro-library in such a way that micro-colonies of the replicate library are in substantially the same pattern as the micro-colonies of the original ("mother cell") micro-library. Additionally, the present invention relates to maintaining, screening, examining and manipulating micro-libraries.
Making and screening banks of cells is an important aspect of many molecular and cellular biological techniques important in laboratory research and in the clinical laboratory setting. For instance, two fundamental endeavors in biotechnology which involve cell banks are gene cloning and production of monoclonal antibodies.
Gone cloning is a tool which is key to nearly all areas of biological research. For instance, it is crucial to producing many pharmaceutical compounds, particularly those of proteinaceous nature. Gone cloning makes possible the elucidation of the genetic basis of many diseases, and it will be a central feature of genetic therapies which may address a variety of dysfunctions caused by alterations in gene expression.
Cloning genes and other DNA segments can be an arduous task, however, and realizing the full potential of gene cloning techniques, a goal of the effort to sequence the entire human genome, is beyond the reach of current technology. For instance, the human genome consists of approximately 3.times.10.sup.9 base pairs per haploid complement. (For an authoritative discussion of genomic and cDNA cloning, including a description of cloning vectors and mathematical considerations in genomic and cDNA cloning, see Sambrook et al., MOLECULAR CLONING, A LABORATORY MANUAL, Second Edition, Vol. 1-3 (Cold Spring Harbor Laboratory, 1989)). An average plasmid cloning vector is able to accommodate perhaps 6.times.10.sup.3 base pairs of heterologous DNA and still be able efficiently to transform recipient cells using conventional techniques.
Accordingly, 5.times.10.sup.5 recombinant plasmids are required to accommodate an entire human genome. Moreover, because cloning involves inherently probablistic procedures, about 3.times.10.sup.6 such plasmids are required for a 99% chance that all of the plasmids together contain every sequence in the human genome. Even this estimate fails to account for loss of sequences that deleteriously affect transformed-cell growth, something that can be a particular problem when regimes of competitive growth are employed in making, propagating or amplifying a library. In any event, the screening of three million plasmid clones is a formidable undertaking that requires a considerable investment of effort, time and money.
Non-plasmid cloning vectors are available which can accommodate larger segments of DNA. Bacteriophage lambda-derived vectors, for instance, can accommodate approximately as much as 2.5.times.10.sup.4 base pairs of heterologous DNA, thus requiring about one million such recombinants for a complete human library at the 99% confidence level. See, for instance, pages 270-271 of Maniatis et al., MOLECULAR CLONING, A LABORATORY MANUAL (Cold Spring Harbor Laboratory, 1982). Hybrid plasmid-phage vectors known as cosmids accommodate even larger inserts, reducing to perhaps one-half million the number of recombinants needed to provide a complete human genome at a 99% confidence level.
Yeast artificial chromosome (YAC) vectors accommodate even larger inserts, potentially as large as 1.times.10.sup.6 base pairs. By the use of YAC vectors, therefore, it is possible at the 99% confidence level to contain a human genome in approximately 5.times.10.sup.4 YAC recombinants. Even such YAC libraries, however, are cumbersome and inefficient to manipulate and time-consuming and expensive to use. For instance, YAC libraries are often propagated in the wells of microtiter plates, for convenience in maintaining the clones of the library in isolation from each other to prevent cross-contamination. For a human YAC library, about 300 to 500 96-well plates are needed to provide the 30,000 to 50,000 wells necessary to maintain separately each clone in the library. See, for instance, MacMurray et al., Nuc. Acids Res., 19: 385 (1991). Screening 300 microtiter plates one time requires considerable effort, repetitive screening, as required by chromosome walking techniques, for example, requires resources beyond the reach of many laboratories.
Thus, even with the most efficient cloning systems currently available each manipulation of a human library represents a considerable undertaking that is expensive of effort, resources, time and money. In consequence, a variety of interesting and important experiments are rendered impractical or can be undertaken only by marshalling large-scale support.
Another problem in isolating particular clones from libraries is the necessity to clonally purify or ("clone-out") positives. Generally, libraries of complex genomes constructed in plasmid and phage-derived libraries are spread or plated at high densities to minimize the number of petri-plates needed to accommodate the library. Colonies and plaques which are identified as being of interest, therefore, cannot readily be separated from nearby colonies or plaques that generally are not of interest. For each positive clone this necessitates multiple rounds of low-density screening and re-isolation to obtain a homogeneously pure isolate. Rescreening positives slows down and adds to the cost of cloning experiments, and can seriously impede or render impractical experiments where libraries are being screened for the presence of several different sequences or where "walking experiments" that require repeated sequential screenings are being carried out.
The necessity of rescreening positives is mitigated or altogether avoided when libraries are initially spread at low densities and then individual clones are picked into individual cultures in separated wells in an array. Ordered libraries are extraordinarily labor-intensive to construct and it is generally impractical to construct ordered libraries for more than a few tens of thousands of clones. (For a discussion of the difficulties even of an automated procedure for making an ordered library see Uber et al., BioTechniques 11: 642). Furthermore, such low density libraries are cumbersome to manipulate and expensive to maintain and screen.
For instance, ordered human YAC libraries of have been made by transferring individual clones from an initial library spread at low density into individual microtiter plate wells. Even when semi-automated, however, producing such libraries requires extraordinary effort. Moreover, a human YAC library, even an ordered one, must still contain 30,000 to more than 60,000 clones in 300 to over 600 microtiter dishes, and is very difficult to manipulate and expensive to screen.
Similar considerations apply to a variety of other endeavors where populations of cells are produced, propagated and screened to identify those few cells and/or their progeny that possess a desired characteristic. Such an endeavor is the production of monoclonal antibodies, which usually entails fusing immortalized non-secreting B-cells, such as MOPC cells, with the lymphocytes of spleens from animals inoculated with an antigen of interest. The procedure results in clonally derived hybridoma cell lines, each obtained from the fusion of one MOPC cell and one splenic lymphocyte. The bank of hybridomas produced by each fusion experiment must be screened to identify the few clones that produce an antibody having a desired specificity. Often many fusions must be carried out, and a great number of cell lines must be screened, to obtain a hybridoma cell-line that produces an antibody with desired properties. Since it is necessary to propagate each hybridoma cell-line, or a limited mixture of cell lines, in a different microtiter-dish-well, these experiments also often require numerous micro-titer dishes. Consequently, the experiments are cumbersome and time-consuming to carry out, and incur considerable costs, similar to those required for large-scale cloning projects.
Among other applications where large numbers of different cells must be screened that encounter similar problems are toxicology, mutagenicity and carcinogenicity assays.
In genetic toxicology assays, for instance, indicator cells are exposed to a compound, such as a potential pharmaceutical agent, or to an environmental agent, to assess the agent's toxicity, mutagenicity or carcinogenicity, for instance. A common "indication" method is to score a genetic marker, such as reversion of a mutation that prevents growth on a defined medium. In this type of assay, mutational events are counted by the number of colonies that appear on defined, growth-preventing medium after exposure to a potentially mutagenic agent.
These assays typically proceed by exposing the agent to large numbers of indicator cells. Many separate experiments of this type are carried out to determine dose-response curves with sufficient accuracy to measure mutagenic activity. Each point on a dose-response curve may represent the average of multiple assays, each requiring a separate cell-culture. Thus, establishing a single dose-response curve for an agent on a single type of indicator cell may involve many separate cultures for each type of indicator cell.
In addition to requiring many separate cultures, assays of this type can also be especially time consuming in many cases because the indicator cells may be enfeebled and grow slowly, even after a reversion event, and the assays require macroscopic colonies for scoring. Furthermore, small colonies produced by slow-growing mutants may be overlooked systematically when only large colonies are scored, which results in underestimating mutagenic effects, and misrepresenting test results. Accordingly, assays of this type also are logistically hindered by the need to manipulate large numbers of cell-cultures and to screen large numbers of macroscopic colonies, and they are expensive and time-consuming to perform.